Emc Corporation

Emc IP Holding Company LLC

Avinash Kallat

Robert A Thibault

Michael Healey

Naser Marmash

J Michael Dunbar

Michael Craig Fishman

Brandon L Paul

Mark Botello

Andrew Kniager

Stephen Lawrence Scaringella

Almir D Davis

Stephen D MacArthur

Steven T McClure

Christopher S MacLellan

Steven R Chalmer

Brett D Niver

William F Baxter

Amnon Izhar

Michael Sgrosso

Nicholas Paluzzi

Krzysztof Dobecki

John Phinney

Philip M Sailer

Techniques for performing de-duplication for data blocks in a computer storage environment. At least one chunking/hashing unit receives input data from a source and processes it to output data blocks and content addresses for them. In one aspect, the chunking/hashing unit outputs all blocks without checking to see whether any is a duplicate of a block previously stored on the storage environment. In another aspect, each data block is processed by one of a plurality of distributed object addressable storage (OAS) devices that each is selected to process data blocks having content addresses with a particular range. The OAS devices determine whether each received data block is a duplicate of another previously stored on the computer storage environment, and when it is not, stores the data block.

Method and apparatus for block level data de-duplication

Techniques for performing de-duplication for data blocks in a computer storage environment. At least one chunking/hashing unit receives input data from a source and processes it to output data blocks and content addresses for them. In one aspect, the chunking/hashing unit outputs all blocks without checking to see whether any is a duplicate of a block previously stored on the storage environment. In another aspect, each data block is processed by one of a plurality of distributed object addressable storage (OAS) devices that each is selected to process data blocks having content addresses with a particular range. The OAS devices determine whether each received to data block is a duplicate of another previously stored on the computer storage environment, and when it is not, stores the data block.

A data transfer system includes a PCI Express transaction layer having an input for serially receiving posted and non-posted request packets and completion packets; an application layer coupled to the PCI Express transaction layer for receiving posted and non-posted request packets and completion packets from the PCI Express transaction layer; a first transmission interface coupling the application layer to the PCI Express transaction layer; and a second transmission interface coupling the application layer to the PCI Express transaction layer. The PCI Express transaction layer transmits posted and non-posted request packets to the application layer over the first transmission interface and transmits completion packets to the application layer over the second transmission interface.

Method and system for processing packet transfers

Described is an end-to-end broadcast-based messaging technique used in controlling message flow in a data storage system. Each node stores flow control state information about all the nodes which is used in determining whether to send a data transmission to a receiving node. The flow control state information includes an indicator as to whether each node is receiving incoming data transmissions. If a node is not receiving incoming data transmissions, the flow control state information also includes an associated expiration time. Data transmissions are resumed to a receiving node based on the earlier of a sending node determining that the expiration time has lapsed, or receiving a control message from the receiving node explicitly turning on data transmissions. Each node maintains and updates its local copy of the flow control state information in accordance with control messages sent by each node to turn on and off data transmissions. Each node sends out control messages in accordance with predetermined threshold levels taking into account hardware and/or software resources for message buffering.

End-to-end broadcast based flow control in a switch fabric

An interrupt processing system having an interrupt holding registers, each corresponding to a different class of interrupts. A write queue posts servicing required by the interrupt holding registers. An interrupt vector register has bit positions corresponding to different classes of interrupts. A read queue has inputs coupled to the plurality of interrupt holding registers and to the interrupt vector register. Detection logic is coupled between an arbiter, fed by the write and read queues, and a processor for: (a) indicating when an interrupt has passed from the write arbiter to the processor; (b) detecting the interrupt class of such passed interrupt; (c) enabling the one of the bit positions corresponding to the detected interrupt class in the interrupt vector register to store a state indicating the servicing requirement for such detected class of interrupt; and (d) wherein the data stored in the interrupt vector register is passed to the processor through the read queue and the arbiter selector.

Interrupt processing system

A system and method for determining data integrity as such data passes through a FIFO. A generator is provided for appending a bit in a predetermined bit location in each packet pushed into the FIFO in response clock signals. The appended bit is a function of the information pushed into the FIFO. A checker is provided for providing an indication of the information integrity in response to bits produced at an output of the FIFO in the predetermined bit location. In one embodiment, the generator is a parity generator and the checker is a parity checker. In one embodiment, during an initial test mode, one parity type is introduced into the FIFO by the parity generator and the opposite parity type is checked at the output of the FIFO by the parity checker to determine whether the parity checker is able to produce parity error signals. In another embodiment, the generator is a packet delimiter generator and the checker is a packet delimiter checker. In another embodiment, the generator is a frame delimiter generator and the checker is a frame delimiter checker.

First-in/first-out (FIFO) information protection and error detection method and apparatus

Messages signaling interrupt (MSI) processing system

An interrupt controller enables multiple CPUs to control access to an increased number of interrupts. Each of a plurality of CPUs is able to block interrupts written to the interrupt controller at multiple levels. First, each CPU is able to block interrupts at the interrupt level. In other words, a CPU is able to block one or more individual interrupt requests from I/O devices from being sent to that CPU. Second, each CPU is able to block interrupts from one or more entire MSI interrupt registers from being sent to that CPU. The interrupt controller is fully programmable by the CPUs in software and thus is very flexible, as the priority of interrupts can be controlled by the CPUs according to the requirements of the CPUs based on the various operational demands of the CPUs. Any of 512 possible interrupt requests are capable of being routed to any particular one CPU, any combination of the CPUs or to all of the CPUs.

System for controlling interrupts between input/output devices and central processing units

A system for maintaining data coherency. The system includes a plurality of processors. A plurality of resources is also included. One portion of the resources is sharable with the plurality of processors and each one of the other ones of the resources being dedicated to a predetermined one of the processors. The system also includes a plurality of buffers. Each one of the buffers is associated with a corresponding one of the plurality of processors. Each one of the buffers is adapted to successively store information presented thereto in successive locations of such one of the buffers. The information includes requests from the corresponding one of the processor. The system includes a logic section responsive to each one of the requests provided by the plurality of processors. The logic section produces indicia indicating whether or not such one of the requests is a request for an operation with one of the sharable resources. The logic section inserts the indicia into a succeeding one of the locations of the buffers associated with such one of the processor providing such one of the requests. The logic section also simultaneously inserts such indicia into a next succeeding available location of the other ones of the plurality of buffers. The logic section inhibits execution of requests in the buffers stored in locations thereof succeeding the location having stored therein any such inserted indicia.

Multi-processor system having data coherency

A system interface includes a plurality of first directors, a plurality of second directors, a data transfer section and a message network. The data transfer section includes a cache memory. The cache memory is coupled to the plurality of first and second directors. The messaging network operates independently of the data transfer section and such network is coupled to the plurality of first directors and the plurality of second directors. The first and second directors control data transfer between the first directors and the second directors in response to messages passing between the first directors and the second directors through the messaging network to facilitate data transfer between first directors and the second directors. The data passes through the cache memory in the data transfer section. A method for operating a data storage system adapted to transfer data between a host computer/server and a bank of disk drives. The method includes transferring messages through a messaging network with the data being transferred between the host computer/server and the bank of disk drives through a cache memory, such message network being independent of the cache memory.

Data storage system having separate data transfer section and message network with status register

Data storage system having separate data transfer section and message network with data pipe DMA

A direct memory access (DMA) transmitter includes: (a) a data register; and (b) a transmitter state machine. Requested data at an address provided by a source is read from the random access memory then transferred for storage in the data register. The central processing unit also sends a control signal to the transmit state machine. The control signal indicates to the transmit state machine whether the read data is a most recent copy of the requested data in random access memory or whether the most recent copy of the requested data is still resident in the local cache memory. In response to the control signal, if the most recent data is in the local cache memory, the transmit state machine inhibits the data that was read from random access memory and now stored in data register from passing to the transmitter output. Transmit state machine then performs a second data transfer request at the same address, the second requested data being transferred from the local cache memory to the random access memory. The transmit state machine reads the second requested data from the random access memory. The second requested data is the most recent data available in the random access memory. The transmit state machine then stores such second requested data into the data register, such stored second requested data then being transferred to the transmitter output.

Direct memory access (DMA) transmitter

A direct memory access (DMA) receiver adapted to receive data from a source, such data to be written into a random access memory is provided. The random access memory and DMA receiver being coupled are to a central processing unit by a bus. The central processing unit is coupled to a local cache memory. The source of such data provides an address for the data, such address being the location the random access memory where the data is to be stored. The DMA receiver includes an address register, a first data register and a duplicate data register. The duplicate register has an input coupled to an output of the first data register. A selector is provided having a pair of inputs, one being coupled to the output of the first data register and another one of the pair of inputs being coupled to an output of the duplicate data register. The selector couples one of the pair of inputs to an output thereof selectively in accordance with a select signal. A state machine is included in the DMA receiver. The state machine is responsive to a receive write enable signal from the source. In response to such signal, the state machine loads the address into the address register and loads the data into the first data register transferring the data onto the bus for storage in the random access memory at the address loaded into the address register. The state machine also checks to determine whether the data corresponding to the address being written into the random access memory is currently stored in the local cache memory. The state machine also transfers the data in the first data register into the duplicate data register. If the data is located in the local cache memory, the state machine waits for the central processing unit to copy data in the local cache memory to the addresses in the random access memory, produces the select signal to couple the data in the duplicate data register to the bus, and overwrites the data in the local cache memory with the data provided on the bus by the duplicate data register.

Growing community of inventors

Marlborough, MA, United States of America

Avinash Kallat